Product Thinking in Data & AI

AI in Production

Slides: https://drive.google.com/file/d/1udC3-LAGZPaHgmxu3zyiMPTYeNLZlIY1/view?usp=drive_link

Adam Becker [00:00:05]: And without further ado, I want to introduce our next and first guest, Stuart. Let's see. Stuart, are you around?

Stuart Winter-Tear [00:00:13]: I am.

Adam Becker [00:00:14]: Okay. Nice. Very good to have you.

Stuart Winter-Tear [00:00:17]: Thank you.

Adam Becker [00:00:18]: You do seem like a lovely chap. You mentioned that you'll be a lovely chap. App. You are here to discuss exactly one of these things that I've just been speaking about. The key to unlocking the full value of AI being not just in technology, if not perhaps like not entirely in technology, but in product. This is something that I've believed in for a very long time. But I have a feeling that you will be much more articulate in being able to explain this in your perspective. So the floor is yours.

Adam Becker [00:00:48]: I'll be back in ten minutes. And you need to share your screen. Right, I think you got it.

Stuart Winter-Tear [00:00:55]: Yep, I think so. Can you see? Yep. Okay. Hi, everybody.

Adam Becker [00:00:59]: Thank you.

Stuart Winter-Tear [00:01:00]: I'm Stuart Wintertier, head of AI product at Geneos. We haven't got long, so if you want to find out more about me, please do go on LinkedIn and find me. The first thing is, why are we talking about connecting AI to value? I can't be alone in seeing the plethora of articles that are rising up lamenting the lack of ROI on data and AI projects. And I want to make it clear from the outset, I am not blaming data teams for this. Many of them have been set up for failure. They're siloed and cut off from the wider business and from the market, treated as waiters at a table. And I just want you to know that I do not blame data teams for this. My thesis is, just because we can research and build it doesn't mean they'll come.

Stuart Winter-Tear [00:01:47]: I think the startup world is evident of that. The failures in the startup world. The core part of my thesis is product strategy, not technology, is the key to winning with AI. And I'll expand on that. Of course, focus on the problem we're trying to solve. We love to solutionize, but sometimes we just need to kind of dig down on what the actual problem is and business outcomes and customer value first. So how do we go about thinking about this? This is the product quartet. I love simple frameworks.

Stuart Winter-Tear [00:02:17]: It's a system of asking ourselves questions to try to drill down on this. Now, desirability really focuses in on the customer side. It focuses in on the problem side. Will this create value for the customer? Will this solve the problem? And will it create demand in the market? Viability is the business side. Are we in sync with the wider business strategy? Can we create defensible modes. And note, defensible mode is not just technological. It can be pricing, packaging, it can be the story that we tell, branding and so on. So don't just think about technological and is there a measurable ROI for the business? Feasibility is going to be more common to us because we're looking at the technological advantage.

Stuart Winter-Tear [00:02:59]: Can we build support, scale? What we're proposing to build, and the new one is datability. Do we have access to the data and is the quality good enough? It used to be the product triad, now it's the product quartet. Moving on to the product mindset in general. Generative AI looking at opportunities, I say the flaws of the opportunities. I don't just mean the flaws in processes and workflows, but also the flaws in generative AI itself. Large language models, things like security, privacy, governance, hallucinations, those sorts of things that will be solved through the use of small language models, working in concert with large language models. A lot of opportunities there. Data is not the new oil.

Stuart Winter-Tear [00:03:39]: Oil is depleted when it's used, whereas data is regenerative. In other words, it doesn't get used up. It actually starts to gain more data as we create data products, which will lead to more innovation and better products. I believe in democratizing AI innovation from the bottom up, from the frontline workers. They know their workflows rather than penalizing the user. Generative AI, we need to surface it up in a safe way so that we understand how they're using it and why they're using it. And it's a good opportunity for us to do discovery and opportunities to augment humans far greater than automating existing tasks. I would much rather focus in on getting my employees to be ten times more productive with generative AI than look to offload 10% of them.

Stuart Winter-Tear [00:04:28]: I'm sure you'd agree. Looking at innovation, the western linear mind loves a straight line between here and there. Innovation is not like that. Very often the best we can do is the next right thing based on the incomplete data that we have. And with that comes a lot of ambiguity and unknown. So large project plans, large grand duos, product roadmaps, don't work in true innovation. And that doesn't make everybody comfortable. I was boring my wife in the car on one of my product monologues, and I was talking about products that are solutions in search of problems.

Stuart Winter-Tear [00:05:05]: And my wife asked me why this was the case, and I said, the problem is often not the problem. That's the problem. What I mean by that is we need to spend time in the problem space, what is the problem? What is the context of the problem? What's the problem under the problem? Is the problem the real problem? Really understand that before we jump to building and solutionizing is very key. But having said all that, we still need to know the grand vision of where we're heading. If we're heading nowhere, we'll find it. Challenges in generative AI hallucinations were featured, not above. That's a hot take. We love the nature, the unexpected things that happen with generative AI and the emerging capabilities.

Stuart Winter-Tear [00:05:47]: But we have to be very careful in our use cases, especially in critical environments. But we're not in that deterministic, rules based world anymore. We're in the world of probabilities. And that's a good thing. 90% of pilots are not going to production. This is a disconnect again between data and the wider business. We must have the two connected together in order for the pilots to be shown to generate value and get buy in. And I just want to mention the hype.

Stuart Winter-Tear [00:06:15]: I love this quote from Amara's law. We will see, and we are seeing a degree of dissatisfaction and disappointment with the ROI on generative AI that will not lead to an AI winter. And we cannot underestimate the massive impact that this will have, which I believe will be huge. Just a quick note on trust and transparency from a product perspective. I love this slide. It's not mine, but I love the way it's broken down into technology and humans. Architecture and communication explainability on the architecture side, very, very important. Not just for us as builders, but also if possible, to have explainability for our customers as well.

Stuart Winter-Tear [00:06:53]: Why did the model make this decision? Feedback loops incredibly important that we can give our customers the opportunity to feedback to us. When our models go wrong, it helps them in participation and gets rid of frustration and connected with that is education, transparency, letting our customers know what our models can and cannot do. At Ux, one of my favorite parts, this will be transformed from second or third to first place. Multimodal and hyper personalisation is going to take center stage and moving from app to app to get tasks done will be over. I believe that we'll see a single pane of glass regenerative AI orchestrating the back and I look forward to it strategy have a clear vision for how we deliver measurable business and customer value. So mature KPIs we can prove our RoI and get buy in. And I say it over and over again. The human API.

Stuart Winter-Tear [00:07:47]: This came from MIT and McKinsey. The connector and the translator. This is the role that connects data to the wider business. I believe the product is well positioned to be able to take this role, to be honest. But they have to be able to speak the language of both business and data. Not an easy task, but a vital thing to have in place. In my opinion, product strategy must ladder up to the business strategy and goals. I've seen it when it's disconnected.

Stuart Winter-Tear [00:08:17]: And the products, the prototypes that do not deliver value to the business and they don't get buy in, the two must be connected together and this is really important. Monetize incrementally in the near term, quick wins, quick value, show the business, get the buy in, then we can start working on the big plans and the big goals that we've got has to be done. Technology is the last, Mark. This is what I'm saying. Product strategy is the first. That's my thesis. That's the core of it. And I love this quote from Andy Sutton, because it doesn't come from the product world, it comes from the analytics world.

Stuart Winter-Tear [00:08:55]: I've been doing this analytics thing for over 20 years now and I thought I'd seen it all and got the t shirt. But the product approaching the enormous value it can unlock genuinely excites me, and me too. That's why I'm doing this talk. There's so much value to unlock. I know this has been very, very rapid. I probably should have taken the 20 minutes talk rather than the ten minute. I do speak about this and these themes a lot on LinkedIn, so please do find me there. And I'm always happy to do a more expanded talk if there's any interest in it.

Stuart Winter-Tear [00:09:25]: So thank you so much for listening. And I was determined not to be the one to go over time. So I've done it. I've done it.

Adam Becker [00:09:32]: No, you're good, Stewart. Thank you very much. I do want to hear more. It feels to me like, if only I had spoken to you like 15 years ago, you would have saved me maybe like a decade of my life.

Stuart Winter-Tear [00:09:45]: Yeah, I know what you mean. I do know what you mean, definitely.

Adam Becker [00:09:48]: Unfortunately, you get to really appreciate all of these things that you just said after a lot of scars and a lot of.

Stuart Winter-Tear [00:09:55]: A lot of pain. A lot of pain. That's how it's learned. I know.

Adam Becker [00:09:58]: Tell me. And I also feel like I have a new tattoo idea. I'm going to get it on my forehead. The problem is often not the problem. That's the problem.

Stuart Winter-Tear [00:10:09]: My wife understood what I meant. She's the only person. It was amazing.

Adam Becker [00:10:15]: Brilliant. Stuart, thank you very much for joining us. And we'll follow you on LinkedIn. Absolutely.